-
Notifications
You must be signed in to change notification settings - Fork 113
/
test_l2rpn_idf_2023.py
241 lines (204 loc) · 11.8 KB
/
test_l2rpn_idf_2023.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
# Copyright (c) 2023, RTE (https://www.rte-france.com)
# See AUTHORS.txt
# This Source Code Form is subject to the terms of the Mozilla Public License, version 2.0.
# If a copy of the Mozilla Public License, version 2.0 was not distributed with this file,
# you can obtain one at http://mozilla.org/MPL/2.0/.
# SPDX-License-Identifier: MPL-2.0
# This file is part of Grid2Op, Grid2Op a testbed platform to model sequential decision making in power systems.
import grid2op
from grid2op.gym_compat import GymEnv, BoxGymActSpace, BoxGymObsSpace, DiscreteActSpace, MultiDiscreteActSpace
from grid2op.l2rpn_utils import ActionIDF2023, ObservationIDF2023
from grid2op.Opponent import GeometricOpponentMultiArea
from grid2op.Reward import AlertReward
import unittest
import warnings
import numpy as np
import pdb
class TestL2RPNIDF2023Tester(unittest.TestCase):
def setUp(self) -> None:
with warnings.catch_warnings():
warnings.filterwarnings("ignore")
# this needs to be tested with pandapower backend
self.env = grid2op.make("l2rpn_idf_2023", test=True)
self.env.seed(0)
self.env.set_id(0)
def tearDown(self) -> None:
self.env.close()
return super().tearDown()
def legal_action_2subs(self):
act12 = self.env.action_space({"set_bus": {"substations_id": [(3, (1, 2, 1, 2, 1)), (33, (1, 2, 1, 2, 1, 2))]}})
act23 = self.env.action_space({"set_bus": {"substations_id": [(33, (1, 2, 1, 2, 1, 2)), (67, (1, 2, 1, 2))]}})
act13 = self.env.action_space({"set_bus": {"substations_id": [(3, (1, 2, 1, 2, 1)), (67, (1, 2, 1, 2))]}})
obs, reward, done, info = self.env.step(act12)
assert not info["is_illegal"]
self.env.reset()
obs, reward, done, info = self.env.step(act13)
assert not info["is_illegal"]
self.env.reset()
obs, reward, done, info = self.env.step(act23)
assert not info["is_illegal"]
self.env.reset()
def test_illegal_action_2subs(self):
# illegal actions
act11 = self.env.action_space({"set_bus": {"substations_id": [(3, (1, 2, 1, 2, 1)), (4, (1, 2, 1, 2, 1))]}})
act22 = self.env.action_space({"set_bus": {"substations_id": [(33, (1, 2, 1, 2, 1, 2)), (36, (1, 2, 1, 2, 1, 2))]}})
act33 = self.env.action_space({"set_bus": {"substations_id": [(67, (1, 2, 1, 2)), (68, (1, 2, 1, 2, 1, 2, 1)) ]}})
obs, reward, done, info = self.env.step(act11)
assert info["is_illegal"]
self.env.reset()
obs, reward, done, info = self.env.step(act22)
assert info["is_illegal"]
self.env.reset()
obs, reward, done, info = self.env.step(act33)
assert info["is_illegal"]
self.env.reset()
def test_legal_action_2lines(self):
# legal actions
act12 = self.env.action_space({"set_line_status": [(0, -1), (110, -1)]})
act23 = self.env.action_space({"set_line_status": [(110, -1), (3, -1)]})
act13 = self.env.action_space({"set_line_status": [(0, -1), (3, -1)]})
obs, reward, done, info = self.env.step(act12)
assert not info["is_illegal"]
self.env.reset()
obs, reward, done, info = self.env.step(act13)
assert not info["is_illegal"]
self.env.reset()
obs, reward, done, info = self.env.step(act23)
assert not info["is_illegal"]
self.env.reset()
def test_other_rewards(self):
assert "alert" in self.env.other_rewards
assert isinstance(self.env.other_rewards["alert"].template_reward, AlertReward)
def test_illegal_action_2lines(self):
# illegal actions
act11 = self.env.action_space({"set_line_status": [(0, -1), (1, -1)]})
act22 = self.env.action_space({"set_line_status": [(110, -1), (111, -1)]})
act33 = self.env.action_space({"set_line_status": [(3, -1), (7, -1)]})
obs, reward, done, info = self.env.step(act11)
assert info["is_illegal"]
self.env.reset()
obs, reward, done, info = self.env.step(act22)
assert info["is_illegal"]
self.env.reset()
obs, reward, done, info = self.env.step(act33)
assert info["is_illegal"]
self.env.reset()
def test_to_gym(self):
env_gym = GymEnv(self.env)
for k in ["active_alert",
"attack_under_alert",
"time_since_last_alert",
"alert_duration",
"total_number_of_alert",
"time_since_last_attack",
"was_alert_used_after_attack"]:
assert k in env_gym.observation_space.spaces, f"missing key {k} in obs space"
assert "raise_alert" in env_gym.action_space.spaces, f"missing key raise_alert in act space"
with warnings.catch_warnings():
warnings.filterwarnings("ignore")
box_act = BoxGymActSpace(self.env.action_space,
attr_to_keep=(
"set_line_status",
"change_line_status",
"set_bus",
"change_bus",
"redispatch",
"set_storage",
"curtail",
"raise_alert",
))
assert box_act.shape[0] == 1543, f'{box_act.shape[0]} vs 1543'
box_act2 = BoxGymActSpace(self.env.action_space)
assert box_act2.shape[0] == 69, f'{box_act2.shape[0]} vs 69'
box_obs = BoxGymObsSpace(self.env.observation_space)
assert box_obs.shape[0] == 5125, f'{box_obs.shape[0]} vs 5125'
disc_act = DiscreteActSpace(self.env.action_space)
assert disc_act.n == 147878, f'{disc_act.n} vs 147878'
multidisc_0 = MultiDiscreteActSpace(self.env.action_space)
assert multidisc_0.shape[0] == 1543, f'{multidisc_0.shape[0]} vs 1543'
multidisc_1 = MultiDiscreteActSpace(self.env.action_space, attr_to_keep=["raise_alert"])
assert multidisc_1.shape[0] == 22, f'{multidisc_1.shape[0]} vs 22'
multidisc_2 = MultiDiscreteActSpace(self.env.action_space, attr_to_keep=["sub_set_bus"])
assert multidisc_2.shape[0] == 118, f'{multidisc_2.shape[0]} vs 118'
assert np.array_equal(multidisc_2.nvec, [ 4, 4, 8, 10, 17, 4, 4, 14, 3,
1, 58, 254, 4, 4, 242, 4, 64, 6,
30, 4, 4, 4, 30, 8, 8, 4, 58,
4, 4, 9, 8, 32, 4, 30, 4, 4,
33, 5, 4, 30, 4, 114, 4, 4, 14,
14, 8, 4, 65506, 4, 8, 4, 4, 126,
14, 498, 4, 4, 506, 14, 16, 58, 3,
5, 16, 62, 4, 9, 64, 122, 5, 4,
1, 4, 32, 6, 1010, 4, 4, 1018, 3,
8, 14, 4, 62, 4, 1, 4, 64, 26,
4, 254, 4, 32, 4, 62, 4, 4, 4,
2034, 4, 4, 16, 14, 62, 8, 6, 4,
4, 16, 1, 1, 10, 4, 4, 1, 1,
4])
def test_forecast_env(self):
obs = self.env.reset()
for_env = obs.get_forecast_env()
assert for_env.max_episode_duration() == 13 # 12 + 1
def test_correct_action_observation(self):
"""test the observation and action class"""
obs = self.env.reset()
act = self.env.action_space()
assert isinstance(obs, ObservationIDF2023)
assert isinstance(act, ActionIDF2023)
assert obs.dim_alerts == 22
assert act.dim_alerts == 22
assert np.all(act.alertable_line_ids == [106, 93, 88, 162, 68, 117, 180, 160, 136, 141, 131, 121, 125,
126, 110, 154, 81, 43, 33, 37, 62, 61])
def test_maintenance_attack(self):
# test the attacks
assert isinstance(self.env._oppSpace.opponent, GeometricOpponentMultiArea)
opp = self.env._oppSpace.opponent
assert len(opp.list_opponents) == 3
line_attacked = []
for sub_opp in opp.list_opponents:
line_attacked += sub_opp._lines_ids
assert np.all(line_attacked == [106, 93, 88, 162, 68, 117, 180, 160, 136, 141, 131, 121, 125,
126, 110, 154, 81, 43, 33, 37, 62, 61])
# test the maintenance
time_series = self.env.chronics_handler.real_data.data
time_series.line_to_maintenance
assert time_series.line_to_maintenance == {'21_22_93', '93_95_43', '80_79_175', '88_91_33', '41_48_131', '62_58_180',
'26_31_106', '62_63_160', '44_45_126', '48_53_141', '34_35_110',
'74_117_81', '12_14_68', '39_41_121', '54_58_154', '17_18_88',
'91_92_37', '4_10_162', '43_44_125', '48_50_136', '29_37_117'}
def test_was_alert_used_after_attack(self):
self.env.seed(0)
obs = self.env.reset()
for i in range(13):
obs, reward, done, info = self.env.step(self.env.action_space())
act = self.env.action_space()
obs, reward, done, info = self.env.step(act) # an attack at this step
assert info["opponent_attack_line"] is not None
# count 12 steps
for i in range(12):
obs, reward, done, info = self.env.step(self.env.action_space())
assert obs.was_alert_used_after_attack[0] == 1
def test_alertreward_counted_only_once_per_attack(self):
self.env.seed(0)
obs = self.env.reset()
for i in range(13):
obs, reward, done, info = self.env.step(self.env.action_space())
act = self.env.action_space()
obs, reward, done, info = self.env.step(act) # an attack at this step
assert info["opponent_attack_line"] is not None
for i in range(11):
obs, reward, done, info = self.env.step(self.env.action_space())
assert info["rewards"]["alert"] == 0, f"error for step {i}"
assert obs.was_alert_used_after_attack[0] == 0
obs, reward, done, info = self.env.step(self.env.action_space()) # end of the time window
assert obs.was_alert_used_after_attack[0] == 1
assert info["rewards"]["alert"] != 0
for i in range(15):
obs, reward, done, info = self.env.step(self.env.action_space())
assert info["rewards"]["alert"] == 0, f"error for step {i}"
assert obs.was_alert_used_after_attack[0] == 0, f"error for step {i}"
def do_not_run_oom_error_test_act_space_alert(self):
# this crashed
all_act = self.env.action_space.get_all_unitary_alert(self.env.action_space)
# bug is fixed but OOM error !
if __name__ == '__main__':
unittest.main()